Re: Problems with large swap partitions

[Thomas Schwinge]

Hello!

On Thu, May 28, 2009 at 07:02:11PM +0200, Marc Dequènes (Duck) wrote:
> First, i don't know if Hurd can handle several swap partitions, so i  
> deactivated the initial one, just to be sure my result is not biased.

This needs investigation.  But I may have used that in the past, I think.


> At boot, or run manually, swapon result in a segfault. Samuel  
> suggested to try 1.7GB, which is known to work, as used on buildds  
> currently. It failed in the same way. I attached the rpctrace, but i  
> don't know what to do now. Could someone help ?

> [...]
> swapon:  114->io_write_request ("swapon: " -1) = 0 8
> /dev/hd2s1: Linux 2.0 swap signature, 1781736k swap-space (-1650988k bad, 
> 1650992k wasted at end) 114->io_write_request ("/dev/hd2s1: Linux 2.0 swap 
> signature, 1781736k swap-space (-1650988k bad, 1650992k wasted at end)" -1) = 
> 0 97
>  114->io_write_request ("

This very much looks like the following scenario -- which I experienced
earlier today (v0 vs. v1 swap signature).

    sh-3.2# mkswap.linux /dev/hd1
    mkswap.linux: warning: truncating swap area to 130752 KiB
    Setting up swapspace version 0, size = 130748 KiB
    sh-3.2# swapon -a            
    swapon: /dev/hd1: Linux 2.0 swap signature, 524280k swap-space (-393532k 
bad, 393536k wasted at end)
    Segmentation fault
    sh-3.2# mkswap.linux -v1 /dev/hd1
    Setting up swapspace version 1, size = 524284 KiB
    no label, UUID=6248abc2-64c0-11de-809c-411e4c248c5a
    sh-3.2# swapon -a
    swapon: /dev/hd1: Linux 2.2 swap signature v1, 524284k swap-space (excludes 
8k at end of partition)

Since when is (a) mkswap to be spelled mkswap.linux (only on Hurd
systems?), and since when does one (b) have to explicitly specify the
newer v1 format (failing Linux kernel version autodetection; has it
always been like this?)?


Regards,
 Thomas

signature.asc
Description: Digital signature